WO2013004700A1 - Method for providing target point candidates for selecting a target point - Google Patents

Abstract

The invention relates to a method for providing target point candidates (22, 23, 26) forming a candidate set for selecting a target point from the candidate set by means of a geodetic measuring device, wherein the measuring device has a sighting unit that defines a sighting direction and a camera oriented substantially in the sighting direction. The measuring device is coarsely oriented toward the target point, and an image is recorded in the sighting direction. Furthermore, a search process for certain target object candidates in the recorded image is performed by means of image processing, wherein the search process is performed on the basis of predefined models and wherein at least one respective point representing the target object candidate is associated with each of the target object candidates as a target point candidate (22, 23, 26). In addition, the target point candidates (22, 23, 26) are associated with a candidate set, and respective weight values, in particular probabilities, are derived according to at least one value of a predetermined target point property of the target point candidates (22, 23, 26) and associated with the target point candidates (22, 23, 26). Further, the target point candidates (22, 23, 26) from the candidate set are each provided together with respective information representing the weight value associated with the target point candidate.

Description

 Method of providing destination candidates for

Select a destination

The invention relates to a method for providing target point candidates for selecting a target point from a candidate set with a surveying device after

The preamble of claim 1 and an inventive surveying device for carrying out the method according to

Claim 12.

Numerous geodetic surveying instruments have been known since ancient times for measuring a target point. In this case, spatial standard data are taken in the direction or angle and usually also the distance from a measuring device to the target point to be measured, and in particular the absolute position of the measuring device together with possibly

recorded existing reference points.

Well-known examples of such geodesic

Surveying equipment provide theodolite, tachymeter and

Total station, which is also referred to as an electronic tachymeter or computer tachymeter. A geodetic measuring device of the prior art is described for example in the publication EP 1 686 350. Such devices have elektrosensische angle and possibly distance measuring function, which allow a direction and distance determination to a selected destination. The angle or distance variables are doing inside

Determined reference system of the device and must be linked for an absolute position determination possibly with an external reference system.

In many geodetic applications, a measurement of points is done by there specially designed target objects to be placed. These usually consist of a pole with a reflector (eg an all-round prism) for

Definition of the measuring section or measuring point. In such surveying tasks, a number of data, instructions, language and other information between the target object - in particular a hand-held data acquisition device on the part of the target object - and central measuring device are transmitted to control the measurement process as well as for the definition or registration of measurement parameters. Examples of such data are the identification of the target object (type of prism used), inclination of the pole, height of the

Reflectors over ground, reflector constants or measured values such as temperature or air pressure. This information or situation-dependent parameters are required in order to enable a high-precision alignment and measurement of the measuring point defined by the pole with prism.

Modern total stations have microprocessors for digital processing and storage of acquired measurement data. The devices generally have a compact and integrated design, with mostly coaxial

Distance measuring elements as well as computing, control and

Storage units are present in a device. Depending on the expansion stage of the total station can also be a motorization of the aiming or sighting device and - in the case of the use of retroreflectors (such as

 Round prism) as target objects - means for automatic target search and tracking to be integrated. As a human-machine interface, the total station can do one

electronic display control unit - generally a microprocessor processing unit with electronic

Data storage means - with display and input means, such as a keyboard, have. The display control unit will be fed to the electrosensory detected measurement data, so that the position of the target point by the display control unit can be determined, visually displayed and stored. Total stations known from the prior art can furthermore have a radio data interface for setting up a radio link to external peripheral components, such as a handheld data acquisition device, which can be configured in particular as a data logger or field computer. For sighting or targeting of the target to be measured have generic geodetic surveying a telescopic sight, such as an optical telescope, as

Sighting on. The scope is generally about a vertical standing axis and about a horizontal

Tilting axis rotatable relative to a base of the measuring device, so that the telescope can be aligned by pivoting and tilting on the point to be measured. Modern devices can additively to the optical view channel one in the

Riflescope integrated and, for example coaxially or in parallel aligned camera to capture an image, the captured image in particular as a live image on the display of the display control unit and / or on a display of the remote control used

Peripheral devices - such as of the data logger - can be displayed. The optics of the sighting device can in this case have a manual focus-for example a set screw for changing the position of a focusing optics-or have an autofocus, the changing of the focus position being e.g. by servomotors

he follows. For example, such a sighting device of a geodetic surveying device is described in EP 2 219 011. Automatic focusing devices for Riflescopes geodetic devices are for example from the

 DE 197 107 22, DE 199 267 06 or DE 199 495 80 known.

The optical system or the optical vision channel of

Sighting device usually contains an objective lens group, an image reversing system, a focusing optics, a

Reticule for creating a reticule and a

Eyepiece, e.g. arranged in this order from the object side. The position of the focusing lens group is adjusted depending on the object distance so that a sharp object image is formed on the graticule arranged in the focusing plane. This can then be viewed through the eyepiece or e.g. be detected by a coaxially arranged camera. By way of example, the construction of riflescopes of geodesic devices of the generic type is shown in the publications EP 1 081 459 or EP 1 662 278.

Due to the mostly both visual channel as well as for

Measurements to be used together beam path require such devices whose technical design in the construction of a telescope with specialized, high-precision and complex-to-produce optics. Furthermore, for the coaxial electronic distance measurement, an additional separate transmission and reception channel and an additional image plane for the wavelength of the distance meter

intended .

Since target objects (eg, the most commonly used for geodesic purposes plumb bobs with target such as an all-round prism) using the sighting despite the often provided 30-fold optical magnification with the naked eye enough can be precisely targeted (ie non-geodetic accuracy requirements accordingly), have common surveying equipment now a standard

automatic tracking function for target reflectors (ATR: "Automatic Target

 This is commonly done by another separate ATR light source - e.g., a multi-mode fiber output emitting optical radiation with a wavelength in the range of 850 nm - and more specifically, for these

Wavelength sensitive ATR detector (for example, CCD area sensor) additionally integrated in the telescope.

As part of the ATR fine-tuning function, the ATR measuring beam is emitted in the direction of the optical target axis of the sighting device, this is retro-reflected at the prism and the reflected beam is detected by the ATR sensor. Depending on the deviation of the alignment of the optical target axis from the prism, the impact position of the reflected radiation on the ATR sensor deviates from a central sensor surface position (ie, the reflection spot of the ATR measurement beam retro-reflected on the prism on the ATR surface sensor is not centered of the ATR area sensor, and thus does not encounter a target position that has been determined, for example, by calibration as that corresponding to the optical target axis position).

If this is the case, the alignment of the sighting device is motorized slightly adjusted so that the prism retro-reflected ATR measuring beam high-precision in the center of the sensor surface on the ATR area sensor

(that is, the horizontal and vertical angles of the sighting device are iteratively changed and adjusted until the center of the reflexing spot coincides with the target Position on the ATR area sensor coincides).

Alternatively, a residual deviation between impact point of the retro-reflected ATR measurement beam on the ATR area sensor and the center of the sensor area may also be

calculated and converted into an angle which is added to the - detected by the angle sensors - solid angle, in which the target axis shows, is added accordingly. That is, the space angle to the target point can in this case also be derived from the spatial angle of the target axis detected using the angle sensors and a storage of the detected ATR measurement beam reflex from the sensor center (ie from that central point on the ATR sensor). where the target axis is mapped).

Thereby, the achievable accuracy in the alignment of the optical target axis on the prism can be significantly increased over a manual cross-hair targeting and based on sheer eye size. to

Ensuring the functioning of the automatic

Targeting based on evaluation of the position of the

Reflecting spots of the prism retro-reflected ATR measuring beam on the ATR surface sensor, it is necessary to align the sighting device at least approximately on the target reflector prior to operation start, that the retro-reflected ATR measuring beam also impinges on the ATR surface sensor. For this, e.g. previously one

manual targeting of the target reflector based on

Eye size done or an automatic coarse target function ^{¬ be} executed.

The manual, rough sighting of the target object can

on the user side, by placing the target object on a user display of the display control unit directly on the surveying device or on the display of a separate display Peripheral device (eg data logger as a remote control) is considered and targeted. Often, however, this is still done by looking at the target through the eyepiece of the riflescope (ie the sighting device), as in the

practical use - e.g. in sunlight - a displayed display image on the display control unit or the datalogger may be insufficiently visible.

In addition to the ATR fine-tuning function, in a similar manner and using the same ATR components (such as ATRL light source and ATR detector), an automatic ATR can also be used

 Be provided targeting functionality. After ATR fine-graining (i.e., after the sighting device is aimed at the target such that the center of the ATR measurement radiation reflex spot coincides with the target position corresponding to the target axis)

ATR surface sensor coincides) then the targeting device can continue to track such movements of the target "live" and correspondingly fast that the center of the ATR measurement radiation reflex spot continues to remain as accurate as possible and always at the target position on the ATR area sensor It is often said that the target is "locked in". Problems can arise in this case if the target moves so rapidly and quickly that it disappears from the field of view of the ATR detector (i.e., no ATR reflected at the target).

Measuring radiation hits more on the ATR surface sensor).

For example, is in EP 2 141 450 a surveying device with automatic aiming for a

retroreflective target as well as with automatic

Target tracking functionality. In order to keep the target in the "locked" state even with fast and jerky movements and not out of the Viewing area of the fine target detector to lose, is proposed in parallel by a (in the visible

Wavelength range sensitive) camera to record images of the target and to follow movements of the target (or movements of objects moving together with the target) by means of image processing and thereby in the case of a loss of the target from the "locked" state, the retrieval and reconnection of the Furthermore, methods are known from the state of the art, whereby, with the aid of an image and image processing (in particular on the basis of a position of a target object determined in the captured image and dependent on a

known or thereby captured image recording direction) the solid angles are determined by the measuring device to the target object. Such methods are e.g. in WO 2005/026767 or in WO 2006/053837.

In addition, in the European patent application

No. 10168771.3 a motorized and automatic

Aligning a targeting device of a surveying device with a target point of a (retroreflectorless) target. The target (or a set of

different target marks) is already known and has an appropriate for the indication of the target point on the target structure / shape (pattern, shape, etc.).

 By means of an evaluation unit, the target mark is detected on the basis of the pattern in a captured image and the position of the imaged target mark in the image is determined with high precision. Depending on this particular position of the target or one aligned with the target

In the picture, the targeting device can now be precisely aligned. A disadvantage of prior art surveying equipment is, on the one hand, the given requirement of manual fine alignment of the targeting device with a target point. This may have a limiting effect on the accuracy in the measurement of the solid angles, since this process is dependent on the care and ability of the user. Furthermore, the manual, accurate targeting of a target point requires a certain amount of time, which has a negative effect on the productivity of the user. On the other hand, automatic fine-tuning is included

Surveying devices according to the prior art

defined goals limited. That's how it works

described ATR fine-target function only on cooperative targets, such. Prisms, not natural

Target points. Other methods of automatic fine-tuning only work on defined patterns, e.g. on

Targets whose shapes are known to the system and not to objects that have some variability in their shape, e.g. Steeples. It is therefore an object of the present invention

 To provide a method with which a selection of a desired destination with a high

 Automation level, less complicated, faster and more reliable feasible. Another object of the invention is variability in selection

possible target object types, in particular with regard to their shape, to take into account. A specific object of the invention is, in the selection of the desired

Target point attributes attributed to the target point, in particular to support the selection,

to be considered. Another task is that Providing a geodetic surveying device for such a method.

These tasks are achieved through the realization of

characterizing features of the independent claims. Features that the invention in alternative or

educate in an advantageous manner, are the dependent

To claim.

According to the invention, a method for providing target point candidates together with in each case an indicator for a weighting of the candidates or a

Target point probability provided.

In this case, after a coarse alignment of a targeting unit, in particular of a telescope of a surveying device with an integrated camera, an image is taken which, based on methods of digital image processing, with regard to possible target objects or

 Target object candidate is analyzed. From the target objects or target object candidates, target points or

Target point candidates derived. The determined

Destination point candidates, along with an associated probability, may be considered as the user-desired destination point, e.g. be provided on a display. From the set of provided candidates, the system can automatically get a point as desired

Select destination or ambiguity one

predefined number of candidates, in particular those with the highest probability, offer this to a user for manual selection. After the automatic or manual selection, based on the image information, the deviation of the targeting unit from the direction to the selected target point can be calculated, and the surveying apparatus by means provided for pivoting drives on the

Target point to be aligned.

The method according to the invention for providing target point candidates forming a candidate set for selecting a target point from the candidate set is carried out with a geodetic surveying device, wherein the surveying device has a targeting unit defining a targeting unit and a camera oriented essentially in the targeting direction. The surveying device is roughly aligned to the target point and an image in the target direction is detected.

The camera of the surveying device can have a second beam path and thus, e.g. a detection direction

which is not coaxial or absolutely parallel to the beam path of the targeting unit. For example, the camera and its optics have an enlargement factor smaller than that of the targeting unit, e.g. an overview image to receive orientation, and is therefore offset, in particular below, arranged to a target optics. Thus, the orientation of the camera does not have to match the orientation of the targeting unit

but the camera may be substantially aimed in the targeting direction, i.e. for example, such that a targetable by the targeting unit

Target point in the detection range of the camera is.

Furthermore, a search for certain

Target object candidates in the captured image using

Image processing, in particular by means of a

Classification method, wherein the search is performed on the basis of predefined models and the target object candidates each at least one of the respective target object candidates representative point is assigned as a target point candidate. In addition, the destination candidates become one

Assigned candidate quantity and the target point candidates each a weight value, in particular one

Probability, depending on at least one

Expression derived from a predetermined target point property of the target point candidate and assigned. Further, the target point candidates of the candidate set are respectively provided together with information representing the weight value associated with the respective target point candidate.

In particular, at least one target object candidate

which implicitly also provides a target point associated with the object, in particular for targeting that point. Thus, e.g. one

Users or the system information is offered about which target points and / or target objects were recognized in each case with what probability as a target point in question, and in addition object-based and / or point-based a corresponding selection are made. The provision of target objects thus corresponds within the meaning of the invention to a simultaneous provision of a target point associated with the target object and can thus also be understood as providing target points. Thus, providing a destination or destination candidates can be realized both by individual points and by defined structures, for example objects.

Thus, a search is made for target object candidates in the captured image, the image regarding the occurrence of representatives of predefined target object classes

is examined. Become one or more in the picture Target object candidates are found, for these the

corresponding target points determined or assigned and these based on defined weight factors

in terms of a probability or a

Weight value evaluates the extent to which each candidate is the target point desired by the user. After the selection has been made, it is possible, in accordance with the selected target point candidate, to determine an angle offset from the current target direction and to determine the alignment e.g. a telescope of a total station can be corrected via a motorization according to the filing.

A weight value is understood here to be a variable which indicates to what extent a target point candidate is the desired target point. This weight value can also be understood as a probability, wherein it can be stated with what probability a

Target point candidate comes as target point in question.

In other words, the weight value is a measure of relevance, for example, one

Relevance value of information as desired

Destination to be eligible for a respective

Target point candidate represents. In particular, an impact parameter can also be understood as a weight value, the impact parameter representing a scalable parameter for the extent to which a target point candidate represents the desired target point.

In other words, with respect to the weight value, it can be said to what extent a value associated with the target point candidate reflects a target point significance, i. a significance for that indicates that

Target point candidate represents the desired target point. The determination or assignment of the target points based on the target objects can be rule-based, ie if a representative of a target class is found in the image, the target points can be defined as intersections of lines or as centers of circles or ellipses. For example, for a representative of the target object class

"Steeple" a target point as the intersection of the bars (in the captured image recognized as lines) of the cross and another target point as the center of the knob (detected in the captured image as a circle) These rules for the derivation of the target points can the respective

Target class assigned and stored on the surveying device.

With such a method, a user is allowed to choose a desired one in a measurement environment

Destination point from a lot of provided

 Destination point candidates. Alternatively, the target point can be automatically selected in the process. As a result, after a rough direction of a

Surveying system automatically a target point can be selected and the system automatically aligned.

If automatic target point selection due to

Ambiguities can not be achieved, the user can e.g. the most likely destination candidates are offered for manual selection. This makes it possible to carry out the fine-leveling, which requires extensive manual coarse adjustment, automatically and to increase the productivity of a measuring process.

To provide the target point candidates, an image of a measurement environment is detected, wherein when capturing the image, the surveying device with the camera, such as a total station or a theodolite, is aligned such that the desired target point is roughly targeted and thus the Detection direction of the camera is essentially aligned to this point and thus this is in the field of view of the camera. The captured image is examined for the appearance of target object candidates, for which numerous image processing methods, in particular classification methods, are known. On the one hand, features that can be summarized in a feature vector (eg Haar wavelet or HOG features) can be extracted from an image detail. This feature vector can be mapped into the feature space and assigned to a target feature class depending on the location in that space. The selection of the image section can take place through a window of defined size, which is successively, for example, pushed or rasterized line by line over the entire image (sliding window).

Furthermore, to define the image sections, the image can be viewed with regard to the appearance of distinctive structures, the so-called "interest points", for example by means of Förstner

Operator, Harris-Laplace detector and / or Harris corner and edge detector. Thus, prominent structures can be localized in the image whose environments can in turn define a specific image section, on the basis of which the feature vector is calculated.

Such "interest points" may, for example, also be provided by a subsystem, for example an ATR unit (automatic

 Target detection) are provided, wherein by means of an additional beam source, the positions e.g.

reflective goals and can be targeted. Subsequent image analysis of the environment of an "ATR interest point" can thus improve the robustness of the ATR function with respect to a distinction between ATR targets and non-ATR targets For example, the image environment is examined for structures that can represent a retroreflector.

As a further variant for the definition of the image sections from which the feature vectors are calculated, based on the coordinates of known target points, e.g. of fixed points from official surveying, according to the position and orientation of the camera whose position is calculated in the image. The image section may in this case by a predefined environment around this position

be defined.

For example, each feature vector can be one

Classifier be supplied. By means of the classifier an affiliation of the respective feature vector to a predefined class can be determined and its assignment probability or feature probability, i. the quality of the assignment, to be determined.

Following another method for the detection of target objects in a captured image, this can in turn be searched for "interest points." These "interest points" define the position of the image sections from which

Feature vectors (e.g., SIFT or SURF). As part of the classification, the positions of the

"Interest points" in combination with the calculated

Feature vectors with stored in the system, the

Target object classes corresponding object models

compared and evaluated their affiliation.

Furthermore, the image may also look for geometric primitives, e.g. Circles, ellipses or lines e.g. be searched by Hough transform. Based on the

Neighborhood relationships of the detected primitives can a topology graph of the image are constructed, which in the context of a classification according to different, individual target object classes corresponding and stored

Reference graph structures is searched. From a classification of the target object candidates with respect to their membership of the candidate set, i. whether the object detected in the image is a possible one

Target object or not, can be next to the

Assignment to a target class also one more

Attribution probability, which describes the quality of the respective assignment.

With such a classification can thus not only about a consideration of a recognized possible

Target object to be decided, i. Destination point or not destination point, but also a more detailed

 Subdivision of possible target objects (and the associated target points) into different classes, e.g. Round prism or house edge, done. This class division and the assignment probability can continue for one

Determining the weight value or the probability that the respective target point candidate represents the target point desired by the user, be considered or in the calculation of this probability or

Weighting. The classifier can be trained in the context of the method or can be created on the basis of training images and / or already stored. The classifier can be created on the basis of a set of training images or image sections which represent the various classes of target points sufficiently.

In particular, the training images from a first There is a set of frames that represent real target points and whose target point class is known, and a second set of pictures of similar but

unwanted target points and images of the background of the target points that are assigned to a residual class or non-target point class. These training images can be taken during different times of day and under different measuring and weather conditions in order to produce the most robust evaluation procedure possible. For the training of the classifier can depend on the

Method is extracted from each training image in a predefined image area, a feature vector or a topology graph. A feature vector may consist of a

Combination of different geometric and

consist of statistical descriptors that unambiguously and independently of image size and size

Describe orientation. The feature vectors of all

Training images can be converted into a feature space and there take a certain position according to the characteristics typical for the target class. The feature space can now with the help of one, in particular

machine learning, such as the

Maximum Distance Method (Support Vector Machine), so be divided into areas that each class has one

as large a space as possible without overlapping with another class. The feature space can also be transformed into a higher dimensional space to facilitate learning. In addition, the dimension of the feature space can be reduced, for example by means of principal component analysis, in order to prevent statistically irrelevant information and thus to make the training of the classifier more efficient. In addition, a function can be learned for each class that has a relationship between given

Feature vector and its assignment probability. These functions can be learned, for example, by means of logistic regression.

In particular, geometric primitives can be detected in defined image sections during learning, and from these a topology graph can be constructed. Thus, corresponding reference graph structures can be generated for each target object class and stored for the classification, in particular for a "graph matching".

According to the invention, the search process can be carried out as a function of a rating, in particular by means of classifying, a target object feature of the respective target object candidate, in particular wherein the type and / or environment of the object and / or gray value, gray scale gradient and / or

Histogram of a Zielpunktumgebung and / or geometric primitives and / or their neighborhood relationships are evaluated .. Furthermore, the evaluation of the target feature depending on the target object candidates representing and / or Zielobj ekt-like images, in particular

Image sections, done. Furthermore, one can

Feature expression of the target feature and a

Expression coefficient, in particular where the feature expression and the expression coefficient in

Dependence of a position of a respective feature vector in a feature space are determined, in particular wherein an assignment probability is determined.

In the picture, potential target objects, i. Representatives of one or more target object classes, e.g. by means of a

The above method can be found from these respective target point candidates are derived. Each possible destination can have a weight value or

Probability to be assigned, these being

Probability as a measure of how far the respective target point candidate, in particular by the user, desired target point represents. Such

Rating can be based on predefined weight factors

based.

The weighting factors may e.g. of the

Assignment probability, i. the quality of

 Classification, depending on the target class underlying the target point candidate according to the frequency of their occurrence in one

Be derived from the measurement environment. Furthermore, the

Weight factor depends on the distance of the target point candidate to the image center. In addition, according to the

Selection behavior of a user or current

Measuring a weight factor can be determined.

The target point candidates may then be combined with information that determines their respective one

 Probability - based on the weight factors - represents, be provided. After evaluation, either the most likely candidate, i. the one with the highest weighting, automatically selected by system or, if more than one candidate, the most likely

Candidates to the user for manual selection

to be provided. According to the invention, a certain number of destination candidates can be provided, in particular depending on the respective one

Weight value. The provision of the target point candidates can take place in such a way that the target point candidates, in particular together with the information on the weight values, eg on a display, in particular touch-sensitive

Display, be provided. The user can thus be given the opportunity to select the desired target point from the provided set of possible target points, wherein the information regarding the probability or weighting can be taken into account in support of the selection of the target point. In addition, the

 Target point candidates together with the information the

In relation to the probability or the weight value, an automatic selection of a destination from the

Candidate quantity can be taken. In this case, e.g.

that target point candidate is automatically selected, which in relation to other destination candidates a larger, especially significantly larger,

Weight value is assigned. According to the invention can thus in the context of the method of

Destination point from the provided destination candidates and the information representing the respective target point candidate weight value representing information determined by a user or automatically. In particular, that target point candidate can be automatically determined as the target point, which relative to others

Target point candidate is associated with a greater weight value or a greater probability. Furthermore, the provision of the destination point, the destination point candidate and / or the information which is the weight value associated with the respective destination point candidate,

represented on a display. Furthermore, with the method according to the invention, the target point properties, which form the basis for calculating the probability or the weight value of the

Create target point candidates depend on point properties that can be determined in the captured image. In this case, the target point properties can have those properties or features or they can

 Target point candidates are assigned, which are assigned to the associated target objects. Since a user can try to target a desired target point as precisely as possible (in the center of the picture), the position of a possible target object in the picture may, for example, be weighted

influence. Furthermore, the assignment to a

specific target object class affect the weighting. For example, can be used in forms that have a design of a retroreflector or distinctive objects in one

Measurement environment (for example, summit cross or church tower), a weighting in favor of a higher target point probability done. According to the invention, the target point property of a respective target class object, a position of the

Target point candidates in the image and / or a reflectivity of the respective target object candidate depend.

In addition, the inventive method after the rough alignment of the surveying device automatically

be executed so that, for example, a user only targets a desired destination and as

Result of the targeting an already selected target point gets provided, in particular with the

match the desired destination. In addition, a can

Fine target this target by determining the location of a current orientation of the surveying device for desired destination and realignment of the

Surveyor on the target point manually or automatically done.

Thus, according to the invention, a deposit of a target point defined by the targeting direction to the target point in the image can be determined, and the targeting direction can be determined in such a way

Dependency of the filing be aligned that of

Target point is targeted.

A further aspect of the invention is a geodetic surveying apparatus, in particular total station or theodolite, for the provision of target point candidates for selecting a target point with a targeting unit, in particular a target telescope. The targeting unit is pivotable relative to a base of the surveying apparatus for changing its orientation in two axes, wherein the targeting unit defines a targeting direction and has at least one camera for detecting an image substantially in the targeting direction. Furthermore, an angle measuring functionality for high-precision detection of the alignment of the targeting direction and output means for providing information is provided. In addition, the surveying device has a

Control and processing unit on. In a

Database are models for target object recognition by means of image processing and coefficients for evaluation of target point properties of the target point candidates

saved. Furthermore, the surveying device has a

Search functionality on, within the scope of a

Capture an image when executing the

Search functionality through the control and

Processing unit in the picture after the

Target object candidates depending on one

Degree of matching is sought with the models, in particular by means of a classification method, wherein the target object candidates each have at least one point representing the target object as

Target point candidate is assigned. Furthermore, as part of the search functionality, the destination candidates are assigned to a candidate set and are dependent on the

Coefficients respective weight values, in particular

Probabilities derived to the respective target point candidates and assigned to the target point candidates. In addition, the destination candidates are each associated with a respective one of the respective target point candidate

Weight value representing information, in particular on the output means, provided. Such a surveying device according to the invention can also be so

be formed that a described inventive method is feasible, i. the surveying device can provide the search functionality such that at

Execution of the search functionality, an inventive method described above is performed. The invention also relates to a computer program product with program code, which is stored on a machine-readable carrier, for carrying out a method according to the invention for providing destination candidates for selecting a destination from the destination candidates, in particular if the program is stored on a computer

 Processing unit of an inventive geodetic surveying trained electronic

Data processing unit is executed.

The method according to the invention and the measuring system according to the invention will be described below with reference to concrete examples shown schematically in the drawings

Exemplary embodiments described purely by way of example, whereby further advantages of the invention will be discussed. In detail show:

Fig. La-b an orientation of a targeting of a

 Surveying device to a desired destination point according to the prior art;

Fig. 2 is a schematic flow of the inventive

 Method;

Fig. 3a-c an inventive detection of

 Target point candidates in a captured image using a topology graph;

4a-c show an inventive form of classification of target point candidates recognized in the image by means of feature vectors and their position in the feature space;

Fig. 5 with an inventive surveying device

 captured images with different

 Target object candidates with respective target points;

6 shows an inventive provision of

 Destination point candidates with respective associated information; Fig. 7, a determination of the storage of a desired

 Destination from a current orientation of the survey device.

Figures la and lb show an alignment process for a surveying device according to the prior art. When aiming a desired point 21 of an object 20, here

represented as a spire, the orientation of a targeting unit of a surveying device, e.g. one

Theodolites or a total station, changed so much that the center 12 of a crosshairs 11 indicating the alignment of the targeting direction coincides with the target point 21 (FIG. 1b). In this case, a user can aim the target point 21 by means of a diopter and align the telescope roughly by twisting on the object 20. In a second step, the user can look through the telescope and thereby by turning the fine drives of the

Surveying device to a precise orientation on the

Reach destination point 21. This process can be very useful in practice

In addition, the accuracy of the targeting depends on the abilities of the user.

The alignment can also - depending on the shape of the

Target object - with an automatic targeting function

(Automatic Target Recognition, ATR), which allow accurate automatic alignment of the telescope, e.g. allows for a reflector. The disadvantage here proves that the aiming works only on reflectors and not on natural target points 21. Today, total stations are also available with a "tap and turn" function, whereby a live image can be recorded in the target direction with an integrated camera and displayed on a display - on the surveying device or a controller Select point 21 on which the telescope will automatically align itself

Accuracy of the aiming is in turn limited by the ability of the user to hit the target point 21 with the pen in the right place on the display. Especially for measurements to distant targets and taking into account the resolution limits of the display, this process can be made more difficult.

In addition, methods are known with which a

 Automatically and with high accuracy to a non-reflective target. Here, however, it is assumed that the outer shape of the target object 20 is known, e.g. by

Use of defined target plates attached to the target point 21 for the survey and in the form of a corresponding target pattern on the target

Surveying instrument is stored.

FIG. 2 shows a schematic sequence of a

inventive method. The invention is based on the idea of assisting a user in targeting a desired target point 21 and by using

 Image processing methods to achieve a very accurate targeting of this point 21 from a set of possible target points.

In this case, in a first step 51, the telescope of the surveying device can be roughly aligned by the user on the target point 21. This can be done by turning the telescope manually or by using the "Tap and Turn" function, and then, using a camera integrated in the surveying device, an image can essentially be taken in

Targeting be included 52, which means

Methods of digital image processing is analyzed with respect to possible target objects. In this case, structures recorded in the image can be matched with models that are stored in the surveying device and objects can thus be recognized and recognized as possible target objects 53. For each target object then at least one possible target point is derived 54.

These possible target points are then grouped together, with each element of that set

evaluated in terms of that probability or

55 is weighted 55 that the corresponding point is the target point 21 desired by the user, which is to be targeted automatically by the system if necessary.

The set of possible destination points can then, in a next step 56, together with information,

especially with the determined weighting or

Probability, e.g. be provided to the user on the display or in the surveying device. In this case, the user at similar size

Probabilities or weights, in particular a certain number of candidates - e.g. only those with the highest probabilities or weights - are offered for manual selection.

Depending on the distribution of the probabilities or weights determined in each case to possible destination points, then either the user can manually or the system

automatically select a point from the set of detected and provided points 57, in particular those with a significant probability or weight

After the selection, in a further step 58, the deviation of the selected target point 21 from the current targeting direction in the image can be measured again by methods of digital image processing and, based thereon, correction angles for a realignment of the Targeted. These can be sent to motorized drives for alignment of the telescope and thus serve for a precise alignment 59 of the telescope on the target point 21. FIG. 3 a shows an inventive detection of

possible target object candidates 20,25 in a captured image 10, wherein in the following an inventive

Classification of the possible target candidates 20,25 will be described in conjunction with Figures 3b and 3c.

FIGS. 3b and 3c show an inventive form of the classification of target object candidates 20, 25 on the basis of a topology graph.

The captured image 10 may with regard to possible

Target objects 20, 25 are examined, i. it should be

automatically determine all objects in the image that qualify as candidates for automatic targeting. In the case shown, target points of a

Steeple 20 and a summit cross 25 determined.

This image analysis can be carried out on the basis of classification methods. In a first variant, the image may be generated by geometric primitives, such as e.g. Lines, circles, ellipses are searched. Based on the

Neighborhood relationships of the detected primitives may construct a topology graph 100 (Figure 3b) of the entire image, taking into account the nature, location and combination with adjacent geometric primitives. By the combination of primitives, e.g. Crosses 22, circles 23 or peaks 24 are detected, with a synopsis of these objects a

Church spire 20 can define. The topology graph 100, which is derived from the spire 20 and the summit cross 25 shown in Figure 3a, is made up of the node "church tower" 111 and

The node "Church Tower" consists of a further subdivision of the nodes "Cross" 121, "Knauf" 122 and "Tip" 123, which in turn consist of individual geometrical primitives to find these primitives in the image Methods of digital image processing, such as the Hough Transformation, are used.

In the system, you can access the individual target object classes

corresponding reference graph structures 111, 112 as

Reference models are stored, which are searched in the generated from the image 10 topology graph 100. FIG. 3c shows one of the target object classes

 "Church tower" corresponding reference structure 111 and one of the target object class "summit cross" corresponding

Reference Structure 112.

Within the scope of the classification, the stored graph structures can be searched for in a graph created from the image. Among other things, the method of "graph matching" is known for finding certain topological structures in the graph.

Of the possible detected in Figure 10

Target objects 20,25 become possible

 Target points 22,23,24,26,27 derived. Among these, it is now important to identify those whom the user wants to target. This can be done in a variety of

Target point candidates 22,23,24,26,27 these are subjected to an assessment of that probability, after which it is detected in each individual image Destination point 22,23,24,26,27 is about the desired destination point 21 by the user.

This weighting of the individual target point candidates can be done on the basis of predefined weighting factors, the product of these individual weighting factors being one

 Total weight value for the target point candidate. Since the user with the coarse sighting can already try to hit the desired target point 21, a weighting factor can be the distance of the candidate 22, 23, 24, 26, 27 relative to the current orientation (cross hairs 11) of the candidate

 Be surveying device. That possible target points in the center of the image may receive a higher rating than points located at the edge of the captured image 10.

Another weight factor may be based on the frequency of occurrence of a target 22,23,24,26,27 or

 Target point 21 of a particular target class associated in the surveying practice based. Accordingly

For example, as shown in FIG. 5, round prisms 44 very often represent desired target points 21 and, if recognized as round prisms 44, can thus cause an increased weight value and / or be provided with a corresponding, more weighted coefficient. Summit crosses 41, spire tips 42 and / or specific target point markers 43 may also be recognized on a captured image and may each represent at least one target object candidate, e.g. by a specific training of the spire simultaneously two target points 21st

can be defined. Furthermore, an individual behavior of the user in a manual selection in the determination of the Weight value or relevance in how far a

Target point candidate represents the desired target point. If e.g. the user very often one

Representatives of the class "church tower" and in this selects the center of the pommel, this can be detected by the system and taken into account in future reviews, especially automatically.

In addition, from the settings of the

Surveying instrument or the ongoing measurement program, further criteria for the weighting can be derived. Is e.g. a distance measuring mode set to "reflectorless" may be a possible target point 22,23,24,26,27 of the class

For example, as part of a measurement program, the user may be prompted to measure a polygon point on which a reflector is placed, in which case a possible target point of the "round prism" class would be associated with a "round prism" be assessed according to higher probability. In order to derive the weight value, it is also possible to define a function or an algorithm which can be defined by means of several weighting factors (for example belonging to the

Destination point candidates 22,23,24,26,27 to a particular target class associated with the

Attribution probability to this class and / or

 Distance in the image 10 to the cross hairs) determines a total weight value, which is the

 Destination point candidates 22, 23, 24, 26, 27 can be assigned.

FIGS. 4a and 4b each show an image 10a, 10b of a measurement environment, an image 10a having an ATR unit in FIG. 4a and an image 10b having a camera in FIG. 4b was recorded. In addition, FIG. 4c shows a feature space 32 with that generated from the image 10b

 Feature vectors 86a, 86b, 86c. The ATR captures the image 10a and detects the reflections of a laser beam 84a, 84b, 84c emitted by the ATR. On the basis of the reflections 84a, 84b, 84c, it is difficult to judge in the ATR image alone whether the reflected laser is detected by a 360 ° prism 81, a round prism 82, or another object, e.g. a rearview mirror 83 of a vehicle.

In this case, the positions of the laser spots 84a, 84b and 84c in the context of the invention can act as "interest points" and thus define the position of the image sections 85a, 85b, 85c in the recorded image 10b

Image sections can continue

 Feature vectors 86a, 86b, 86c (Figure 4c) are generated, which are mapped into the feature space 32.

In this feature space 32, regions 87a, 87b are defined which correspond to the individual target object classes. The classification is based on the "position" of the

 Feature vector 86a, 86b, 86c in feature space 32 judges which target object class the corresponding one

Image section 85a, 85b, 85c is assigned. So come here, for example the feature vectors 86a and 86b lie in the regions 87a and 87b, which are e.g. the target object classes

"360 ° prism" and "round prism". Since the feature vector 86c lies outside the defined regions, it can not be assigned to any target object class. In addition to the assignment of the individual

Image sections 85a, 85b, 85c to the target object classes The classification also returns a result

Attribution probability, which determines the quality of the

Assignment describes. This can e.g. are derived from the position of the feature vector 86a, 86b, 86c within a region 87a, 87b. If a vector 86a is located more or less in the center of a region, its

 Assignment probability to be high, e.g. 90%, 86b is a vector at the edge of a region or in the vicinity of another region, the

Assignment probability lower.

Furthermore, the user can be offered the option of even defining target object classes to which the system automatically generates an associated classifier or adapts the existing classifier accordingly. In addition, the classification can be improved by incorporating additional information into the process. As e.g. The size of a round prism is known, from the distance to the object with knowledge of the optical

Image parameters are derived from the size of the prism in Figure 10b. Also, the sizes of other objects, e.g. Summit crosses or pommels on church towers generally move within certain limits. The distance to the object can be determined by means of a distance measuring functionality of the surveying instrument or derived from the position of the focusing lens of the aiming device.

FIG. 6 shows an inventive provision of target point candidates 22, 23, 26. A captured image was scanned accordingly for possible target candidates, including the target point candidates 22,23,26

derived and as possible goals in terms of their

Probability weighted and, in particular, one of the possible targets as a target point, provided with labels 51,52,53, which a selection by

Actuate the corresponding button on the control panel or by tapping the label 51,52,53 with a pen allow provided. If the weight value or

 Probability of a possible target point significantly larger than any other, this can be selected by the system without further interaction of the user. For such a decision as to whether a weight value or one of the probabilities is significantly greater than the other weight values or probabilities, a predefined threshold value can be used in which

Exceeding an automatic selection can be done.

If there are several candidates with similar weights, the selection can be made manually by the user. For this purpose, for example, a predefined

Number of possible destination points - e.g. those with the highest weight value or greatest relevance - marked in the image and e.g. with, a marker or a

Label 51,52,53 be provided. The user can thus by tapping the marker in the display 60, for example, with a pen, or by pressing a button corresponding to the marker or label on the control panel of

Surveying instrument the selection of the destination point

carry out.

Furthermore, a further selection for the abort 54 of the automatic targeting function may be present in the display 60. If this selection is made by the user, the information about the target point candidates 22, 23, 26

hidden and the crosshairs of the target unit

so that the user can manually target a destination in a known manner. In addition, there may be a selection for learning 55 of new target points, where the desired target point can again be manually targeted by the user. By selecting this function, after targeting, the user may be given the opportunity to assign the targeted target point to an existing target class or to a new custom target class

to generate. The system generates one of a section of defined size around the reticle

Feature vector and forms this in an associated

Feature space off. There can be a change

existing class boundaries of the classifier to the

Considering this new target point to be adjusted and thus future target points of this kind as a possible target point candidates 22,23,26 recognize.

FIG. 7 shows a determination of the positional storage of a desired target point 21 of the target object 20 from a current orientation of the surveying device. The filing can be used as a position difference between a

Center 12 of the reticle 11 or between a defined by the orientation of the targeting Anzielpunkt in the image and the desired or selected target point 21 are understood. The storage between the points can be determined by means of image processing in two directions 71, 72, wherein the respective direction deposit 71, 72 can each indicate a distance between the desired destination point 21 and the crosshair center 12.

Depending on the optical parameters of the targeting unit, particularly the telescope, (e.g., focal length) and the integrated camera (e.g., pixel size), the bin may be further adjusted in corrective angles to realign the binoculars

Target unit or the surveying device converted become. These corrections can be made to the motorized ones

Pivoting drives of the target unit are transmitted and thus serve for precise reorientation of the target unit to the desired target point 21, this alignment can be performed automatically controlled.

In addition, the specific tray 71.72 on the

Display 60 are displayed. Thus, a user - as an alternative to automatic alignment - the

Manually realign the target unit by adjusting this fine drive of the target unit so that the position difference is zero. The can

Directional trays 71,72 on the display 60 continuously, in particular continuously, be updated and thus indicate an alignment progress to the user. It is understood that these illustrated figures represent only possible embodiments schematically. According to the invention, the various approaches can also be combined with each other as well as with methods and devices for the aiming of target points and with surveying devices of the prior art.

Claims

claims

A method of providing target point candidates (22,23,24,26,27) forming a candidate set

 Selecting a target point (21) from the candidate set with a geodetic surveying device, the surveying device defining a targeting unit defining a targeting direction and a substantially in

 Targeting directed camera, with

• roughly align the surveying equipment with the

 Target point (21) and

 • Detecting an image (10,10b) in the targeting direction, characterized in that

 • a search for specific ones

 Target object candidates (20,25) in the detected

 Image (10,10b) takes place by means of image processing, in particular by means of a

 Classification method, wherein

 ° The search is performed on the basis of predefined models and

 ° the target object candidates (20,25) each at least one of the respective target object candidates (20,25) representing point as

 Target point candidate (22,23,24,26,27) is assigned,

• the target point candidates (22,23,24,26,27) to the

 Be assigned to candidate quantity,

 To the target point candidates (22, 23, 24, 26, 27) one weight value each, in particular one

 Probability, depending on at least one expression of a predetermined

Target point property of Target point candidates (22,23,24,26,27) is derived and assigned and

 • the target point candidates (22,23,24,26,27) of the

 Each candidate set together with a weight value associated with the respective target point candidate (22, 23, 24, 26, 27)

 Information to be provided.

2. The method according to claim 1

 characterized in that

 the search process as a function of an evaluation, in particular by means of classification, a

 Target object feature of the respective

 Target object candidates (20, 25), in particular type and / or environment of the object and / or gray value,

 Grayscale gradient and / or histogram of a

 Target point environment and / or geometric primitives and / or their neighborhood relationships occurs.

3. The method according to claim 2

 characterized in that

 the evaluation of the target object feature as a function of the target object candidates (20, 25) and / or target-image-like images (10, 10b),

 especially image sections, takes place.

4. The method according to claim 2 or 3

 characterized in that

 a feature expression of the target object feature and an expression coefficient are determined, in particular wherein the feature expression and the

Coefficient of expression as a function of a position of a respective feature vector (86a, 86b, 86c) in one Feature space (32) are determined, in particular wherein an assignment probability is determined.

5. The method according to any one of claims 1 to 4

 characterized in that

 a certain number

 Target point candidate (22,23,24,26,27) is provided, in particular depending on the respective weight value.

Method according to one of claims 1 to 5

 characterized in that

 the destination point (21) from the provided

 Target point candidates (22, 23, 24, 26, 27) and the weight value associated with the respective target point candidates (22, 23, 24, 26, 27)

 Information is determined by a user or automatically.

7. The method according to any one of claims 1 to 6

 characterized in that

 the target point candidate (22,23,24,26,27)

 is automatically determined as a target point (21), which is relative to others

 Target point candidates (22,23,24,26,27) is associated with a larger weight value.

8. The method according to any one of claims 1 to 7

 characterized in that

 providing the destination (21), the

Target point candidates (22, 23, 24, 26, 27) and / or the target point candidates (22, 23, 24, 26, 27) associated weight value representing

Information takes place on a display.

Method according to one of claims 1 to 8

characterized in that

the target point property of a respective one

Target class, a location of the

 Target point candidates (22,23,24,26,27) in the image and / or a reflectivity of the respective

Target object candidates (20,25) depends.

Method according to one of claims 1 to 9

characterized in that

a tray (71, 72) of a target point defined by the targeting direction to the target point (21) in FIG

Image (10,10b) is determined and the targeting direction is aligned in such a manner depending on the tray (71,72) that the target point (21) is targeted.

Method according to one of claims 1 to 10

characterized in that

the method after roughly aligning the

Surveyor is running automatically.

Geodetic surveying device, in particular

Total station or theodolite, for providing target point candidates (22, 23, 24, 26, 27) for selecting a target point (21) with

 A targeting unit, in particular a target telescope, wherein the targeting unit

 ° relative to a base of the surveying equipment to change its orientation in two axes

is pivotable, ° defines a target direction and

 ° at least one camera to detect a

 Image (10, 10b) essentially in the targeting direction,

 • Angular measuring functionality for high-precision detection of the alignment of the targeting direction,

 • Output means for providing information and

A control and processing unit,

characterized in that

 • in a database models to one

 Target object recognition by means of image processing and coefficients for a rating of

 Target point properties of

 Target point candidates (22,23,24,26,27) are stored and

 • the surveying device a search functionality

 in the context of which after detection of an image (10, 10b) when the search functionality is executed by the control and processing unit

 ° in the picture (10,10b) after the

 Target object candidates (20,25) is searched as a function of a degree of agreement with the models, in particular by means of a

 Classification method, wherein the

 Target object candidates (20, 25) are each assigned at least one point representing the target object candidates (20, 25) as a destination point candidate (22, 23, 24, 26, 27),

 The target point candidates (22, 23, 24, 26, 27) are assigned to a candidate set,

 ° depending on the coefficients respective

Weight values, in particular probabilities, to the respective

 Target point candidates (22,23,24,26,27) are derived and associated with the target point candidates (22,23,24,26,27) and

 ° The target point candidates (22,23,24,26,27) each together with the one the respective

 Target point candidates (22,23,24,26,27) associated weight value representing

 Information (51,52,53), especially on the

 Output means are provided.

Geodetic surveying device according to claim 12

characterized in that

the surveying device provides the search functionality such that upon execution of the

Search functionality a procedure according to one of

Claims 1 to 11 is performed.

Computer program product with program code stored on a machine-readable carrier for

Implementation of the method for providing

Destination point candidates (22,23,24,26,27) for selecting a destination (21) according to any one of claims 1 to 11, in particular when the program is on a than

Processing unit of the geodetic surveying device according to one of claims 12 or 13 formed electronic data processing unit is executed.